Re: Why IBMi research effort

Mark,

It sounds to me like you are a very talented and inquisitive young man. I for one can appreciate the concept of hanging probes and connecting wires to your old system attempting to figure things out on your own. I too own DMMs and VOMs, and ociliscopes, and various probes and tools. The hardware guys put out extra security when I get near the hardware booth at COMMON or other shows. I've had a lot of success learning and testing and trying. I've hooked 45 drives to a model 170. I've built a model 270 that was 7 ft tall. I've hung external stuff all over my lab machines. But I do it to show to people how reliable the machine is, to explain that as tested and supported it truly is a rock solid machine and that even pushing twisting and expanding it still runs great. It is over-built to be reliable.

If I may be allowed to interpret your comments you want to be able to tweek the bits. To touch the innards. To root about in the internals. Sad to say IBM i may not be the operating system for you. IBM has intentionally blocked such things in order to enable that rock solid reliability, that work class secure-ability, that ability to run a program from 20 or more years ago unchanged and run it on the very smallest or the vary largest machine completely unchanged. At the same time it also supports the many virtualization capabilities you mentioned, it supports PHP and apache and CGI and Java. I do not understand the issues you mention with Java or PHP. I have installed PHP 25 times and it has worked first time nearly every time. I have dozens of customers running gobs and piles of Java up to and including extremely large SAP implementation.

You mention TN5250 being phased out. I wonder where you get that from? Sure Twinax is gone but that was a 1960s legacy connectivity and while a single mile long cable supporting 7 devices was pretty cool in it's day it's simply not a usable choice today. Today we connect Ethernet to everything it's simpler, faster, cheaper, and more reliable. And yes it supports TN5250 and does so at a blinding speed. IBM Supports TN5250 on most platforms and other vendors support it on nearly everything.

You point out "a million ways to do things' which is what happens when a system is still viable 35 years after it's introduction. Maintaining backward compatibility (something the Windows guys don't even know the definition of) means there will be a lot of options. But just because they are there doesn't mean you have to use them. Use the modern options, pick what works. There is no need to be a guru on every possible component.

True the hobbyist value is small. If you truly want to learn about the current versions of the system check out my offering at www.iDevCloud.com. There you'll be on the current version rather than a version released 6 1/2 years ago (about a century in computer years!)

You claim IBM i is complicated for the sake of being complicated. I don't see that at all. I see elegance and logic. Built in layers to isolate the hardware bits from the software and operating system while leveraging the hardware at the same time.

I'm curious about all those failures in the logs. Are they truly failures (remember you've been sticking probes in the old gal) or are they noticies, observations and such. Every look at the logs in Linux or Windows? I suggest you won't find them devoid of any mystery errors! Yo also comment that the error codes are mysterious and must be searched for to determine their meaning. Perhaps the codes themselves are a bit less than obvious but every message has one! Many Many messages in windows etc are just words so you get to search for text instead. Ill take 7 characters every time!

Every system is different to be sure. IMHO all in all with reliability capability expandability compatibility IBM i on POWER compares favorably to every operating system on there in must respects.

- Larry "DrFranken" Bolhuis

On 8/31/2012 5:54 PM, Mark D wrote:

I'll go ahead and chime in here. I am a relatively new convert to the
IBM i platform. A casino I work at is using it and I've been slowly
gathering steam and learning the system. I am probably a bit younger
than you guys, 30 so I am speaking from the perspective of someone whose
always had cheap software and a fast upgrade cycle around.

The reasons to use the IBM i are obvious and numerous.
From reading inside the as/400 it's obvious the security of the system
is excellent - fine you win this one but most i admins are old school
and leave gaping holes like giving everyone the sst password via email.
Administered well, it is a system to be trusted. Way ahead of its time.
You can run tons of things in LPAR VM's, etc. I can run linux and
windows on the same box. Alright it was cool when it debuted in the 820
days. It's pretty old hat now.
Integrated database - it's pretty cool, very good performance on huge
indexed tables. agreed.
On a whole the system is cool. It was always (way) ahead of it's time.
In general you can do anything you want.

Reasons not to use IBM i:
* Doing anything you want is complicated. It's like trying to fit a
square peg in a round hole. There's no point in doing *anything*
commodity on an i like PHP webserving, java, file serving or anything
like that because it's a colossal waste of money vs getting pc servers.
* TN5250 is being phased out. Ever watch someone work in 5250 vs
navigator? 5250 is THE killer feature of the platform and IBM would
like to kill it. Try to teach an entire building of people to use a
linux console. LOL. Yet somehow IBM i has done it.
* I once read that choosing IBM i is choosing freedom. Sure, freedom to
have your box connect back to IBM without you asking it to, freedom to
display weird error codes that you can then translate instead of just
telling you what's wrong. Freedom to pay massive amounts of money for
software. Freedom to pay WAY more for hardware than it really should
cost. There is nothing "free" about the 820 I bought but at least
replacement disks (which are failing at an alarming rate) are cheap.
* IBM i has little to no hobbyist value. Attempting to legitimately
acquire a nearly useless 10 year old system is difficult to impossible.
Things like *P10 and *P30 make no real sense. An 820 at *P30 runs much
more slowly than a current *P05. I understand IBM wants to sell more
hardware, but they aren't going to sell ANYTHING at my company when I am
CTO later on if kids have no way to dabble with the platform. If the
kids can't work on it, it's a dead end as far as I'm concerned. It's
not a question of if but when.
* Which brings me to my 820 I bought on ebay for $75. It was liberated
from some company who was too lazy to move the 350 pound monster. IBM
still thinks that a V5R4 license on this P30 machine is worth just as
much on a machine they are selling now and won't even talk to me since I
don't have valid PoE. Their death grip on the platform and software is
why I would *never* use that platform as an ISV. Oh but we have MI? In
case you haven't noticed, POSIX gets the job done pretty nicely with
linux as long as you use a sufficiently high level language like C++.
IBM writes the SLIC in C++. What's good for the goose is good for the
gander. Besides as an ISV the planned obsolescence is beneficial, you
sell new versions. Using a platform like linux means less money went to
IBM and more can go to you.
* You're not allowed to say as/400 anymore. Well that's just stupid.
As/400 was a huge instantly recognizable brand with marketing power.
Throwing that away was just stupid. It will always be an as/400 to me.
* IBM i is so powerful. It's amazing. I can pay MANY thousands of
dollars and have the best. For a year or two. No thanks, I'd rather
pay less for commodity hardware and get more of it. FAR more options
from doing business that way.
* I'm doing my best to learn RPG but look this language is a total brain
bender. When you compare it to something like C#.net or C++ it just
doesn't make a whole lot of sense. To me the old school RPG has value
and the latest free form RPG is just silly. Don't emulate a C like
language, just use it, or use real RPG! (maybe sans indicators)
* No way to use SDA screens in anything but RPG as far as I can tell.
Being able to use this stuff from C++ would be a huge boon.
* IBM hides the details in the SLIC. Sorry but this drives me BONKERS.
I want to know what's going on. I am THAT guy with the logic probes
hooked all over my 820 trying to figure out things like where the serial
number is stored.
* For some reason there is still not "Linq to IBM i". Once you go ORM
you don't really want to go back.
* There are a million ways to do things. There is IFS and QDLS. There
is RPG with indicators, then there is modern RPG with procedures, then
there is UBER modern RPG with free form. This is TERRIBLY confusing to
someone new to the platform.
* I can't emulate an i in a VM on my laptop. I can't download the
database so that I can work over a slow/no link on a train or plane. I
*require* *good* connectivity to the server. They could build an IBM i
laptop or nettop even just for developers. But IBM i clearly couldn't
give a crap about NEW developers.
* Basically no integration with windows mobile, windows phone or
iphone. All mobile integration thus far must be done through web
services and client access.
* The most important thing I see is that IBM i is completely insane.
It's complicated for the sake of being complicated and stupid in so many
obvious places. Join 2 tables in sql and if you have a 7 digit decimal
and a 9 digit decimal in another table and link against these two fields
the index won't hit. You have to cast so that the data types match
precisely. Really dumb. Maybe fixed in 6.1 or 7.1 but we're using
v5r4. I've had 6 years of exposure to the platform and a functional
system in my own house and I don't know everything about the platform
yet. Linux on the other hand I downloaded one day and 2 years later I
was administering linux servers professionally (and competently).
* I see right through the legendary reliability. Looking back through
the logs of my 820 there have been a WHOOOOLE lot of failures. I have
an older Compaq dl360 server that has lost one fan in its entire life.
Our 520 at the casino has had it's fair share of issues too. Which by
the way when it goes down is VERY OBVIOUS to everyone since it's a
single point of failure. I'm not the admin but it seems the failover
system never really picks up when these things happen.

As a technologist I'm enamored with the platform. I do think it's truly
great. But I am seriously concerned that it is 100% a dead end because
IBM couldn't care less about attracting... well. ME. And certainly not
24 year old me. IBM should embrace old boxes getting recycled to
individuals and should embrace tinkering. The world of computing now is
a lot different than the 60's 70's and 80's and IBM shouldn't try to
create something that is more of the same. The differences of the
platform are what make it good. Embrace those differences and accept
that people are going to use commodity boxes along side of an i.

Also the performance is never as promised. I ran an HDtach test (on the
IXS) of my 820 against a _12_ drive array and the throughput was not
great at all. I understand that this is a server from 2003 but it is an
$85,000 server from 2003 and it is really not much better than 3 $2000
servers from 2003.

I hope I don't get flamed too bad. But I'm just a younger guy and
that's my impression.

Thanks,
Mark



On 8/31/2012 3:54 PM, DeLong, Eric wrote:

James,

You really lament (resent?) the abandonment of 5250, don't you... Quite frankly, I'd blame the legacy of twinax and 5250 for holding this platform apart from the rest of the datacenter if I had to...